Evolutionary Training of Binary Neural Networks by Evolution Strategy

Hidehiko Okada¹

Section:Research Paper, Product Type: Journal-Paper
Vol.9 , Issue.1 , pp.32-36, Feb-2021

Online published on Feb 28, 2021

Copyright © Hidehiko Okada . This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
 

View this paper at   Google Scholar | DPI Digital Library

XML View     PDF Download

How to Cite this Paper

Abstract :
A problem with deep neural networks is that the memory size for recording a trained model becomes large. A solution to this problem is to make the parameter values binary. A challenge for the binary neural networks is that they cannot be trained by the ordinary gradient-based optimization methods. This paper applies Evolution Strategy (ES), an instance of evolutionary algorithms, to the training of binary neural networks and evaluates its ability. The experimental results with the classification task revealed that ES could well optimize parameter values so that the trained model accurately classify both trained and untrained data, if the hidden layer included sufficient units. As the binary parameter value, {-1,1} was found to be significantly better than {0,1}.

Key-Words / Index Term :
Evolutionary algorithm; Evolution strategy; Neural network; Network quantization; Neuroevolution.

References :
[1] G. E Hinton, R. R. Salakhutdinov, “Reducing the Dimensionality of Data with Neural Networks,” Science, Vol.313, Issue.5786, pp.504-507, 2006.
[2] G. E. Hinton, S. Osindero, Y. W. Teh, “A Fast Learning Algorithm for Deep Belief Nets,” Neural Computation, Vol.18, No.7, pp.1527-1554, 2006.
[3] Y. L. Boureau, Y. L. Cun, “Sparse Feature Learning for Deep Belief Networks,” Advances in Neural Information Processing Systems, pp.1185-1192, 2008.
[4] I. Sutskever, G. E. Hinton, “Deep, Narrow Sigmoid Belief Networks are Universal Approximators,” Neural Computation, Vol.20, No.11, pp.2629-2636, 2008.
[5] Y. Bengio, “Learning Deep Architectures for AI,” Foundations and Trends in Machine Learning, Vol.2, No.1, pp.1-127, 2009.
[6] H. Larochelle, Y. Bengio, J. Louradour, P. Lamblin, “Exploring Strategies for Training Deep Neural Networks,” Journal of Machine Learning Research, Vol.10(Jan), pp.1-40, 2009.
[7] X. Glorot, Y. Bengio, “Understanding the Difficulty of Training Deep Feedforward Neural Networks,” Proceedings of the Thirteenth International Conference on Artificial Intelligence and Statistics, Vol.9, pp.249-256, 2010.
[8] P. Vincent, H. Larochelle, I. Lajoie, Y. Bengio, P. A. Manzagol, “Stacked Denoising Autoencoders: Learning Useful Representations in a Deep Network with a Local Denoising Criterion,” Journal of Machine Learning Research, Vol.11(Dec), pp.3371-3408, 2010.
[9] R. Salakhutdinov, G. Hinton, “An Efficient Learning Procedure for Deep Boltzmann Machines,” Neural Computation, Vol.24, No.8, pp.1967-2006, 2012.
[10] A. Krizhevsky, I. Sutskever, G. Hinton, “ImageNet Classification with Deep Convolutional Neural Networks,” Advances in Neural Information Processing Systems, Vol.25, pp.1097-1105, 2012.
[11] A. Graves, A. Mohamed, G. Hinton, “Speech Recognition with Deep Recurrent Neural Networks”. IEEE International Conference on Acoustics, Speech and Signal Processing, pp.6645-6649, 2013.
[12] Y. Bengio, A. Courville, P. Vincent, “Representation Learning: a Review and New Perspectives,” IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol.35, No.8, pp.1798-1828, 2013.
[13] Y. LeCun, Y. Bengio, G. Hinton, “Deep Learning,” Nature, Vol.521, No.7553, pp.436-444, 2015.
[14] J. Schmidhuber, “Deep Learning in Neural Networks: an Overview,” Neural Networks, Vol.61, pp.85-117, 2015.
[15] S. Zhang, A. E. Choromanska, Y. LeCun, “Deep Learning with Elastic Averaging SGD,” Advances in Neural Information Processing Systems, pp.685-693, 2015.
[16] I. Goodfellow, Y. Bengio, A. Courville, “Deep Learning,” MIT Press, 2016.
[17] M. Courbariaux, Y. Bengio, JP. David, “BinaryConnect: Training Deep Neural Networks with Binary Weights during Propagations,” Proceedings of the 28th International Conference on Neural Information Processing Systems (NIPS’15), pp.3123–3131, 2015.
[18] X. Lin, C. Zhao, W. Pan, “Towards Accurate Binary Convolutional Neural Network,” Proceedings of the 31st International Conference on Neural Information Processing Systems (NIPS’17), pp.344–352, 2017.
[19] H. Qin, R. Gong, X. Liu, X. Bai, J. Song, N. Sebe, “Binary Neural Networks: A Survey,” Pattern Recognition, Vol.105, 107281, 2020.
[20] H.P. Schwefel, “Evolution Strategies: a Family of Non-Linear Optimization Techniques based on Imitating Some Principles of Organic Evolution,” Annals of Operations Research, Vol.1, pp.165–167, 1984.
[21] H.P. Schwefel, “Evolution and Optimum Seeking,” Wiley & Sons, 1995.
[22] H.G. Beyer, H.P. Schwefel, “Evolution Strategies: a Comprehensive Introduction,” Journal Natural Computing, Vol.1, No.1, pp.3–52, 2002.